From ee57646f9ec3e7005b7624dc8586557883238314 Mon Sep 17 00:00:00 2001
From: Dr-Noob <peibolms@gmail.com>
Date: Thu, 5 Aug 2021 15:47:18 +0200
Subject: [PATCH] [v0.98][PPC] Update ppc peak performance taking into account
 slices in POWER9

---
 src/ppc/ppc.c   | 8 ++++++++
 src/ppc/uarch.c | 8 ++++++++
 src/ppc/uarch.h | 1 +
 3 files changed, 17 insertions(+)

diff --git a/src/ppc/ppc.c b/src/ppc/ppc.c
index 66fc397..1a69b5b 100644
--- a/src/ppc/ppc.c
+++ b/src/ppc/ppc.c
@@ -215,6 +215,14 @@ char* get_str_peak_performance(struct cpuInfo* cpu, struct topology* topo, int64
   double flops = topo->physical_cores * topo->sockets * (freq*1000000);
   if(feat->altivec) flops = flops*4;
 
+  // POWER9 has the concept called "slices". Each SMT4 core has two super-slices,
+  // and each super-slice is capable of doing two FLOPS per cycle. In the case of
+  // SMT8, it has 4 super-slices, thus four FLOPS per cycle.
+  if(is_power9(cpu->arch)) {
+    int threads_per_core = topo->logical_cores / topo->physical_cores;
+    flops = flops * (threads_per_core / 2);
+  }
+
   if(flops >= (double)1000000000000.0)
     snprintf(string,size,"%.2f TFLOP/s",flops/1000000000000);
   else if(flops >= 1000000000.0)
diff --git a/src/ppc/uarch.c b/src/ppc/uarch.c
index 4be62f2..61eda1d 100644
--- a/src/ppc/uarch.c
+++ b/src/ppc/uarch.c
@@ -249,6 +249,14 @@ bool has_altivec(struct uarch* arch) {
   }
 }
 
+bool is_power9(struct uarch* arch) {
+  return arch->uarch == UARCH_POWER9      ||
+         arch->uarch == UARCH_POWER9_DD20 ||
+         arch->uarch == UARCH_POWER9_DD21 ||
+         arch->uarch == UARCH_POWER9_DD22 ||
+         arch->uarch == UARCH_POWER9_DD23;
+}
+
 char* get_str_uarch(struct cpuInfo* cpu) {
   return cpu->arch->uarch_str;
 }
diff --git a/src/ppc/uarch.h b/src/ppc/uarch.h
index 7ad20d8..f7c45ce 100644
--- a/src/ppc/uarch.h
+++ b/src/ppc/uarch.h
@@ -8,6 +8,7 @@ struct uarch;
 
 struct uarch* get_uarch_from_pvr(uint32_t pvr);
 bool has_altivec(struct uarch* arch);
+bool is_power9(struct uarch* arch);
 char* get_str_uarch(struct cpuInfo* cpu);
 char* get_str_process(struct cpuInfo* cpu);
 void free_uarch_struct(struct uarch* arch);